Beßt Abstract
Beßt performs efficient non-linear optimization and parameter estimation. Arbitrary functional forms can be optimized or regressed against experimental data. This Microsoft Excel template finds diverse usefulness in science, engineering, medicine, and business.
The application permits flexible choice of optimization methods, such as full 2nd derivative Hessian, or lesser order methods of conjugate gradient, steepest descent, and Gauss-Newton. An original and fast conjugate gradient method is presented, proven to perform remarkably for a wide array of test cases, and useful for problems with large numbers of unknown parameters.
An original line-search is provided and endowed with rule-based artificial intelligence from benchmarking against varieties of industrial and academic cases. This advanced line-search enhances convergence, accuracy, and robustness in cases of high parameter correlation. The template offers a variety of line-search, search accuracy, and optimization strategies to permit necessary rigor when needed, but also simpler, faster, and reduced order strategies for large dimension problems.
A particular attraction of the application is the friendly and familiar Excel interface. The code will check for and build all the necessary formatted workspaces for data and equation input. Building appropriately formatted data and equation spaces is probably the most daunting challenge facing practitioners today who wish to do nonlinear regression and optimization. Confidence intervals are determined for all regressed parameters.
Now, difficult optimization and regression is made easy through the power of this unique application, for users who prefer to work in the familiar environment of MS Excel.
|
